You are here

  • Home
  • Archive
  • Volume 75, Issue 11
  • Anticitrullinated protein antibodies and rheumatoid factor are associated with increased mortality but with different causes of death in patients with rheumatoid arthritis: a longitudinal study in three European cohorts

Article Text

Article menu
Clinical and epidemiological research
Extended report
Anticitrullinated protein antibodies and rheumatoid factor are associated with increased mortality but with different causes of death in patients with rheumatoid arthritis: a longitudinal study in three European cohorts
  1. S Ajeganova1,3,
  2. J H Humphreys2,
  3. M K Verheul3,
  4. H W van Steenbergen3,
  5. J A B van Nies3,
  6. I Hafström1,
  7. B Svensson4,
  8. T W J Huizinga3,
  9. L A Trouw3,
  10. S M M Verstappen2,
  11. A H M van der Helm-van Mil3
  1. 1Rheumatology Unit, Department of Medicine, Karolinska Institutet at Karolinska University Hospital Huddinge, Stockholm, Sweden
  2. 2Arthritis Research UK Centre for Epidemiology, Centre for Musculoskeletal Research, Manchester Academic Health Science Centre, University of Manchester, Manchester, UK
  3. 3Department of Rheumatology, Leiden University Medical Center, Leiden, The Netherlands
  4. 4Department of Clinical Sciences, Section of Rheumatology, Lund University, Lund, Sweden
  1. Correspondence to Dr S Ajeganova, Rheumatology Unit R92, Department of Medicine, Karolinska Institutet at Karolinska University Hospital Huddinge, Stockholm 14186, Sweden; sofia.ajeganova{at}ki.se

Abstract

Objective Patients with rheumatoid arthritis (RA)-related autoantibodies have an increased mortality rate. Different autoantibodies are frequently co-occurring and it is unclear which autoantibodies associate with increased mortality. In addition, association with different causes of death is thus far unexplored. Both questions were addressed in three early RA populations.

Methods 2331 patients with early RA included in Better Anti-Rheumatic Farmaco-Therapy cohort (BARFOT) (n=805), Norfolk Arthritis Register (NOAR) (n=678) and Leiden Early Arthritis Clinic cohort (EAC) (n=848) were studied. The presence of anticitrullinated protein antibodies (ACPA), rheumatoid factor (RF) and anticarbamylated protein (anti-CarP) antibodies was studied in relation to all-cause and cause-specific mortality, obtained from national death registers. Cox proportional hazards regression models (adjusted for age, sex, smoking and inclusion year) were constructed per cohort; data were combined in inverse-weighted meta-analyses.

Results During 26 300 person-years of observation, 29% of BARFOT patients, 30% of NOAR and 18% of EAC patients died, corresponding to mortality rates of 24.9, 21.0 and 20.8 per 1000 person-years. The HR for all-cause mortality (95% CI) was 1.48 (1.22 to 1.79) for ACPA, 1.47 (1.22 to 1.78) for RF and 1.33 (1.11 to 1.60) for anti-CarP. When including all three antibodies in one model, RF was associated with all-cause mortality independent of other autoantibodies, HR 1.30 (1.04 to 1.63). When subsequently stratifying for death cause, ACPA positivity associated with increased cardiovascular death, HR 1.52 (1.04 to 2.21), and RF with increased neoplasm-related death, HR 1.64 (1.02 to 2.62), and respiratory disease-related death, HR 1.71 (1.01 to 2.88).

Conclusions The presence of RF in patients with RA associates with an increased overall mortality rate. Cause-specific mortality rates differed between autoantibodies: ACPA associates with increased cardiovascular death and RF with death related to neoplasm and respiratory disease.

  • Rheumatoid Arthritis
  • Autoantibodies
  • Outcomes research

https://doi.org/10.1136/annrheumdis-2015-208579

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    The presence of autoantibodies is one of the hallmarks of rheumatoid arthritis (RA). Rheumatoid factor (RF) and anticitrullinated protein antibodies (ACPA) are the two autoantibodies that are most characteristic for RA. They have been identified prior to the onset of symptoms of RA and their presence is associated with a more severe disease course, in particular progressive radiographic joint damage.1 ,2 The biological mechanisms by which these antibodies exert their effect are incompletely known, and it is still not clear whether these antibodies have a causal role in the development or the progression of RA. Investigation of the association of these autoantibodies with several disease outcomes might increase the understanding of the role of autoantibodies in RA.

    Several studies have reported increased all-cause mortality in association with RF,3–6 while others in association with ACPA.7 The results of the studies that investigated both autoantibodies are not consistent with some studies reporting an association between increased mortality and the presence of RF, but not ACPA,8 and other studies reporting increased mortality only in ACPA-positive patients.9 Studies that reported increased mortality in the presence of both autoantibodies10 ,11 did not adjust for the fact that the antibodies are frequently present in the same patient, hampering their ability to determine if ACPA and RF are both associated with all-cause mortality, independent of each other.

    The association between RA-related autoantibodies and cause-specific mortality has been infrequently studied in the literature to date. If RF and ACPA are associated with different causes of death, prevalence of which vary in different patient settings, this might explain that some studies observed an increased overall death rate in relation to RF while others found it in relation to ACPA. In patients with RA the incidence of cardiovascular (CV) disease is increased, especially in those who are ACPA positive.12 Based on this, it can be hypothesised that ACPA is also associated with CV death. Furthermore, two small studies13 ,14 from 1970s and 1990s suggested that RF is associated with neoplasm-related death, but ACPA was not determined in these studies. The association of these autoantibodies individually with different causes of death still remains to be explored.

    Anticarbamylated protein (anti-CarP) antibodies have recently been identified in patients with RA, and have been associated with more severe joint destruction.15 In addition, anti-CarP antibodies have been shown to occur years prior to disease onset16–18 and to associate with development of RA in persons with arthralgia.19 It is not yet known whether these autoantibodies are also associated with mortality.

    The aims of this study were to determine the association of the RA-related autoantibodies RF, ACPA and anti-CarP with (1) all-cause mortality and (2) different causes of death, such as CV, neoplasm and respiratory disease attributed causes of death, in three large longitudinal cohorts of patients with early RA.

    Patients and methods

    Patients

    Patients included in the prospective early arthritis cohorts from three countries within Europe, who fulfilled the 1987 criteria for RA20 and with available data on all three ACPA, RF and anti-CarP antibodies were included in this study. Patients had to have been followed for at least 1 year. The local medical ethic committees approved the studies and all patients signed informed consent.

    The Better Anti-rheumatic Farmaco-therapy cohort

    This cohort included 805 out of 1059 patients from secondary care in the southern Sweden with newly diagnosed early (symptom duration ≤1 year) RA. Two hundred and fifty four out of 1059 patients were excluded due to lack of serum for assessment of ACPA and anti-CarP antibodies. Baseline characteristics of the patients included and excluded from the study were similar. Patients were enrolled between 1993 and 1999. For a further cohort description see elsewhere.21 Treatment was started and adjusted during follow-up by the treating rheumatologist in accordance with the current treatment strategies. The majority of the patients included in 1993–1996 were promptly treated with mild disease-modifying anti-rheumatic drug (DMARDs), and methotrexate was early started during the inclusion period 1996–1999.

    The Norfolk Arthritis Register

    Overall, 1212 adults with inflammatory polyarthritis (≥2 swollen joints for ≥4 weeks) were recruited from primary and secondary care in Norfolk UK, between 1990 and 2009. Details of Norfolk Arthritis Register (NOAR) have been described in detail previously.22 Of them, 678 patients met the 1987 criteria for RA, were DMARD-naive and had symptom onset of <2 years at the time of inclusion to the study and were included into the current study. Treatment was initiated and managed during follow-up according to the clinical assessment of the rheumatologists in charge of the patients’ care. Most patients recruited 1990–1995 were started on mild DMARDs, in 1996–1999 methotrexate was introduced, from 2000 onwards early methotrexate and combination therapies come into practice.

    The Leiden Early Arthritis Clinic cohort

    Eight hundred and forty-eight of the total 1124 patients with early RA (symptom duration <2 years) were seen in secondary care and included between 1993 and 2010. Patients recently included (276 out of 1124) lacked anti-CarP data and thus could not participate in the current study. Baseline characteristics of the patients who were included into the study and who were excluded did not differ by age at inclusion, sex, symptom duration, Health Assessment Questionnaire, ACPA and RF positivity. Patients diagnosed in 1993–1995 were initially treated with non-steroidal anti-inflammatory drugs, in 1996–1998 mild DMARDs were started early, from 1999 onwards methotrexate was initiated early.23

    Autoantibody assessment

    Autoantibodies were determined in serum samples collected at baseline for the Better Anti-Rheumatic Farmaco-Therapy cohort (BARFOT) and Leiden Early Arthritis Clinic (EAC) cohorts, and in serum samples obtained within the first year of recruitment in patients of the NOAR cohort.

    The Better Anti-Rheumatic Farmaco-Therapy cohort

    ACPA were tested by the ELISA CCP2 test (Euro-Diagnostica, Malmö, Sweden), a level >25 IU/mL was regarded as positive. IgM RF was measured by agglutination test (Serodia, Fujirebio, Tokyo, Japan), and a level >20 units/mL was defined as positive. The presence of anti-CarP antibodies was determined using in-house ELISA as described previously.15 The cut-off was determined in relation to the anti-CarP antibody reactivity of age- and sex-matched healthy Swedish controls that were assessed in parallel. The cut-off for a positive test in patients was calculated as the mean plus two times SD of the anti-CarP antibody reactivity of the controls. The cut-off was set as a level >452 arbitrary U/mL (aU/mL).

    NOAR

    RF was measured using a particle-enhanced immunoturbidimetric assay where >40 IU/mL was considered positive (Orion Diagnostica, BMG Labtech, Aylesbury, UK). ACPA were measured using the Axis-Shield DIASTAT kit CCP2 test (Axis-Shield, Dundee, UK), positive cut-off >5 U/mL. RF and ACPA were measured in-house. The anti-CarP antibody test was performed as described above, Dutch controls were used and the cut-off for a positive test was set >281 aU/mL.

    The Leiden EAC cohort

    Anti-CCP2 autoantibodies were measured by ELISA (Euro-Diagnostica, Malmö, Sweden), positive cut-off >25 units. IgM RF was determined as a part of routine clinical care. Anti-CarP antibodies were determined as described above, sera of Dutch healthy controls were run in parallel and a level >245 aU/mL was considered positive.

    Outcome assessment

    Outcomes were all-cause and cause-specific mortality. Patients contributed at-risk time from the date of entry into the cohorts. Censoring date occurred at the date of death, embarkation or the date when death data were obtained, whichever came first. Deaths were identified through the national death registers (the Cause of Death Register, Sweden (BARFOT patients, through December 2010), the National Health Service Information Centre, UK (NOAR patients, through December 2014) and the Municipal Personal Records Database, the Netherlands (Leiden EAC patients, through April 2012)). Causes of deaths were coded according to the International Classification of Diseases (ICD), the Ninth and Tenths Revisions (ICD-9, ICD-10). The main cause of death was identified from the death certificates and classified as related to CV diseases (according to ICD-9 codes of 390–459 and ICD-10 codes of I00–I99), neoplasm (140–239 and C00–D49) and respiratory system diseases (460–519 and J00–J99). For the Leiden EAC, all-cause mortality data were available for all patients, but because of regulatory changes in the Netherlands, information on mortality along with causes of death for the EAC patients was accessible only until May 2008, and therefore available for 639 patients.

    Statistical analysis

    Incidence mortality rates with the 95% CI for a Poisson count were presented as events per 1000 person-years at risk. Cox proportional hazards regression models were used with seronegative groups as reference. Adjustment covariates were chosen a priori. Models were constructed with progressive adjustment to show the associations explained by the covariates: unadjusted, adjusted for age and sex and further adjusted for smoking status and year of inclusion as proxy of changed treatment strategy during the extended study period as described previously.23 ,24

    All analyses were first performed in each cohort separately. We anticipated that each individual cohort had insufficient power to make true conclusions particularly in analysis of cause-specific death. Therefore, the effect estimates and SEs of the individual cohort analyses were then combined in inverse-weighted variance meta-analyses. As we assumed that the included cohorts share the true effect, fixed effects models were used if heterogeneity test was statistically non-significant, otherwise random effect models were applied. All significance tests were two-tailed and conducted at the 0.05 level of significance.

    Results

    Patients

    In total, 2331 patients with RA were included: 805 patients from BARFOT, 678 from NOAR and 848 from EAC. Patient characteristics are presented in table 1. The prevalence of different autoantibodies is presented in figure 1. Autoantibodies were often present concurrently, that is, 47% of BARFOT patients, 27% of NOAR and 46% of EAC patients were both RF and ACPA positive. All three autoantibodies were present in 29%, 16% and 35% of the patients in the respective cohorts.

    View this table:
    Table 1

    Baseline characteristics of the patients with RA at inclusion and mortality outcomes during follow-up

    Figure 1
    Figure 1

    Distribution of anticitrullinated protein antibodies (ACPA), rheumatoid factor (RF) and anticarbamylated protein (anti-CarP) antibodies in three cohorts: Better Anti-Rheumatic Farmaco-Therapy cohort (BARFOT), Norfolk Arthritis Register (NOAR) and Leiden Early Arthritis Clinic cohort (EAC).

    The number of person-years of follow-up was 9438 in BARFOT, 9362 in NOAR and 7511 in EAC. During 26 300 person-years of observation, 29% of BARFOT patients, 30% of NOAR and 18% of EAC patients died, corresponding to mortality rates of 24.9, 21.0 and 20.8 per 1000 person-years in the respective cohorts (table 1). In all cohorts, CV death was the most common cause of death (table 1).

    All-cause mortality in relation to ACPA, RF and anti-CarP antibodies

    First, the individual association between each antibody and all-cause mortality was studied. ACPA and RF were associated with an increased mortality risk in all three cohorts, independently of age, sex, smoking and year of inclusion (table 2). There was a trend towards an increased overall mortality risk in association with anti-CarP antibodies in BARFOT and NOAR and a statistically significant association was observed in EAC (table 2).

    View this table:
    Table 2

    HRs (95% CI) for risk of all-cause mortality in association with ACPA, RF and anti-CarP, presented are analyses per individual autoantibody and with all autoantibodies in one analysis

    In the meta-analysis combining data from all three cohorts, ACPA positivity (compared with ACPA negativity) was significantly associated with all-cause mortality, HR (95% CI) 1.48 (1.22 to 1.79). A similar association was observed for RF-positive patients (compared with RF-negative), HR 1.47 (1.22 to 1.78) and anti-CarP-positive patients (compared with anti-CarP-negative), HR 1.33 (1.11 to 1.60) (figure 2).

    Figure 2
    Figure 2

    Meta-analyses of the effect of anticitrullinated protein antibodies (ACPA), rheumatoid factor (RF) and anticarbamylated protein (anti-CarP) on risk of all-cause mortality in patients with rheumatoid arthritis (RA), presented are the results of analyses per individual autoantibody and with all autoantibodies in one analysis. Presented are the HRs (95% CI) of Cox regression models (adjusted for age, sex, smoking and year of inclusion) per cohort (Better Anti-Rheumatic Farmaco-Therapy cohort (BARFOT), Norfolk Arthritis Register (NOAR) and Early Arthritis Clinic cohort (EAC)) and HRs of the meta-analyses with fixed effects p values (random effects p values if indicated). Analyses per individual antibody: ACPA, I2=0.0% p=0.59, fixed effects p≤0.001, random effects p≤0.001; RF, I2=26.1% p=0.26, fixed effects p≤0.001, random effects p≤0.001; anti-CarP, I2=68.9% p=0.041, fixed effects p=0.002, random effects p=0.080; with all autoantibodies in one analysis: ACPA, I2=0.0% p=0.83, fixed effects p=0.15, random effects p=0.15; RF, I2=0.0% p=0.52, fixed effects p=0.024, random effects p=0.024; anti-CarP, I2=61.5% p=0.075, fixed effects p=0.34, random effects p=0.46.

    To determine the association of each of the autoantibodies with all-cause mortality, adjusting for the effect of other antibodies, all three autoantibodies were entered in one Cox regression model (no interaction was found between the autoantibodies). This was first performed in the individual cohorts and showed that RF was associated with a higher mortality rate in BARFOT, HR (95% CI) 1.51 (1.06 to 2.16) and that anti-CarP-antibody positivity was associated with an increased mortality in EAC, HR 1.70 (1.09 to 2.63) (table 2). ACPA was not associated with increased mortality when adjusting for other antibodies in any of the cohorts. Then, meta-analyses on the three cohorts were performed. It showed that RF, independently of presence of other autoantibodies, was associated with an increased overall mortality, HR 1.30 (1.04 to 1.63), but ACPA and anti-CarP antibodies did not show significant associations, HRs 1.20 (0.94 to 1.53) and 1.11 (0.90 to 1.37), respectively (figure 2).

    Cause-specific mortality in relation to ACPA, RF and anti-CarP antibodies

    Then we hypothesised that ACPA, RF and anti-CarP antibodies might be associated with different causes of death; and that these effects might have been undistinguishable when evaluating overall mortality. As the number of deaths per each specific death cause was relatively low, meta-analyses were required to provide sufficient power to detect the autoantibodies’ effect.

    In the meta-analyses evaluating the effect of each autoantibody individually, ACPA, RF and anti-CarP antibodies, compared with the absence of the respective antibody, were all associated with increased CV mortality, the respective HRs (95% CI) were 1.79 (1.37 to 2.34); 1.62 (1.25 to 2.11) and 1.66 (1.27 to 2.16) (see online supplementary table S1 and figure 3). When subsequently entering all three antibodies in one analysis, hence also adjusting for the effect of other two antibodies, only ACPA was associated with increased CV mortality, HR 1.52 (95% CI 1.04 to 2.21) (see online supplementary table S2 and figure 3).

    Figure 3
    Figure 3

    Meta-analyses of the effect of anticitrullinated protein antibodies (ACPA), rheumatoid factor (RF) and anticarbamylated protein (anti-CarP) on risk of cardiovascular-related death, neoplasm-related death and respiratory disease-related death in patients with rheumatoid arthritis (RA), presented are the results of analyses per individual autoantibody and with all autoantibodies in one analysis. Presented are the HRs (95% CI) of Cox regression models (adjusted for age, sex, smoking and year of inclusion) per cohort (Better Anti-Rheumatic Farmaco-Therapy cohort (BARFOT), Norfolk Arthritis Register (NOAR) and Early Arthritis Clinic cohort (EAC)) and HRs of the meta-analyses with fixed effects p values (random effects p values if indicated). Analyses per individual antibody: cardiovascular-related death, ACPA, I2=0.0% p=0.55, fixed effects p≤0.001, random effects p≤0.001; RF, I2=34.6% p=0.22, fixed effects p≤0.001, random effects p=0.012; anti-CarP, I2=26.5% p=0.26, fixed effects p≤0.001, random effects p=0.001; neoplasm-related death, ACPA, I2=0.0% p=0.83, fixed effects p=0.11, random effects p=0.11; RF, I2=39.8% p=0.19, fixed effects p=0.016, random effects p=0.092; anti-CarP, I2=18.5% p=0.29, fixed effects p=0.12, random effects p=0.19; respiratory disease-related death, ACPA, I2=36.5% p=0.21, fixed effects p=0.025, random effects p=0.056; RF, I2=30.6% p=0.24, fixed effects p=0.004, random effects p=0.025, anti-CarP, I2=0.0% p=0.77, fixed effects p=0.076, random effects p=0.076. With all antibodies in one analysis, cardiovascular-related death, ACPA, I2=0.0% p=0.69, fixed effects p=0.029, random effects p=0.029; RF, I2=42.2% p=0.18, fixed effects p=0.32, random effects p=0.69; anti-CarP, I2=78.9% p=0.009, fixed effects p=0.19, random effects p=0.39; neoplasm-related death, ACPA, I2=0.0% p=0.76, fixed effects p=0.99, random effects p=0.99; RF, I2=41.0% p=0.18, fixed effects p=0.041, random effects p=0.19; anti-CarP, I2=58.5% p=0.090, fixed effects p=0.94, random effects p=0.69; respiratory disease-related death, ACPA, I2=0.0% p=0.39, fixed effects p=0.67, random effects p=0.67; RF, I2=0.0% p=0.37, fixed effects p=0.046, random effects p=0.046; anti-CarP, I2=0.0% p=0.87, fixed effects p=0.72, random effects p=0.72.

    Figure 3
    Figure 3

    Continued.

    When evaluating neoplasm-related mortality, no association was observed for ACPA and anti-CarP antibodies, however RF-positive patients, compared with RF-negative, had an increased mortality rate with neoplasm as the underlying cause of death, HR (95% CI) 1.58 (1.09 to 2.10) (see online supplementary table S1 and figure 3). Also, when the other antibodies were included in the analysis, this effect remained, HR 1.64 (1.02 to 2.62) (see online supplementary table S2 and figure 3).

    Finally, the outcome of respiratory disease-related death was studied. In these analyses, ACPA and RF were significantly associated with an increased death rate, the HRs (95% CI) were 1.62 (1.06 to 2.46) and 1.84 (1.21 to 2.80), respectively (see online supplementary table S1 and figure 3). When including all three autoantibodies in one model, only RF was associated with an increased risk of respiratory disease-related death, HR 1.71 (1.01 to 2.88) (see online supplementary table S2 and figure 3).

    Discussion

    This study aimed to increase the understanding of the association between the RA-related autoantibodies RF, ACPA and anti-CarP antibodies and mortality in patients with RA. In three large independent cohorts, we observed that RF-positivity, independently of presence of other autoantibodies, was associated with an increased overall mortality rate in patients with early RA. We did not find support for the independent effect of ACPA or anti-CarP antibody positivity for all-cause mortality. However, we demonstrated that ACPA positivity was independently associated with an increased rate of CV death, whereas RF positivity was independently associated with an increased rate of neoplasm and respiratory disease attributed causes of death.

    Our finding that ACPA-positive patients had an increased CV death compared with ACPA-negative patients is in line with current evidence supporting the prognostic role of ACPA in CV outcome. In this setting, ACPA-positivity has been associated with the development of atherosclerosis,25–27 myocardial dysfunction28 and CV morbidity.12 ,29 Although underlying causal mechanisms of the association are unclear, citrullination of proteins, altering clearing function of macrophages, has been reported to occur within atherosclerotic plaque,30 and ACPA has shown an association with increased pro-thrombotic and pro-oxidative atherogenic immune cell profile.25 Also in the absence of RA, increased levels of ACPAs have been associated with coronary artery disease.31

    Intriguingly, RF was associated with neoplasm-related mortality. Some previous in vitro experiments have suggested a relationship between RF and altered humoral tumour immunity as well as tumour recurrence in cancers.13 ,14 ,32–34 Furthermore, we might consider the possibility that the association between RF and neoplasm-related mortality is not confined to patients with RA. Based on the present data and the fact that prevalence of RF in the general elderly population may be as high as 10%,35 it would be highly interesting to evaluate in future studies if RF is also associated with neoplasm attributed death in persons not suffering from RA.

    The third autoantibody family studied here was anti-CarP antibody. This is the first report investigating anti-CarP antibodies in relation to mortality, and the presence of anti-CarP antibodies was found to be associated with an increased overall mortality. However, this effect might have been driven by the co-occurrence of anti-CarP antibodies with RF or ACPA, as it did not persist after introducing other two autoantibodies into the analyses.

    Carbamylation is like citrullination a post-translational protein modification. Whereas in citrullination arginine is converted by an enzyme into citrulline, in carbamylation cyanate chemically modifies lysines to form homocitrullines that resemble citrulline but one methylene group larger.36 Carbamylated proteins were historically first studied in uraemia patients as urea is in chemical equilibrium with cyanate where it has been proposed to play role in the development of vascular disease via endothelial dysfunction and accumulation of ‘uremic’ carbamylated low-density lipids. Carbamylation has been shown to be present in human atherosclerotic lesions and is proposed to link inflammation, smoking and atherogenesis.37 Higher carbamylation has been reported in association with CV events and all-cause mortality in non-rheumatic patients.37–40 Based on these studies it may be hypothesised that anti-CarP antibody positivity is associated with CV-related death. Although there was a significant association between anti-CarP antibodies and CV attributed death in NOAR, independent of other autoantibodies, this effect was not significant in the meta-analysis.

    A limitation is that the anti-CarP antibody test is not yet fully developed. Healthy control samples randomly selected from the source population have to be considered to determine the positivity cut-offs. Different cut-offs for anti-CarP positivity in the cohorts may have affected our results. Based on our data, the definitive conclusion, whether anti-CarP antibodies are associated with the mortality outcome in RA, cannot be reached.

    Another limitation is that we were not able to test whether titres of ACPA and RF had implication for the mortality outcome, because of the different methods and different cut-offs employed to measure ACPA and RF in the three cohorts. The prevalence of ACPA positivity was lower in the NOAR than in the other two cohorts. Differences in inclusion settings may have accounted for this. Nevertheless, in our view, the similarity of findings in the cohorts and the approach of combining the large data sets sampled in different settings supported the validity of the results.

    A further limitation of the study is that within the cohorts the number of cause-specific deaths was relatively small. To increase the power to detect the effect, we combined data of three cohorts and performed meta-analyses. Still we cannot exclude the possibility of false negative findings (type 2 error) of the analyses of cause-specific mortality. Further studies are required to validate these results in other cohorts of patients with RA. In addition, our study investigated mortality in the patients with RA carrying autoantibodies compared with patients negative for antibodies. It would also be important to estimate the mortality rates of patients with different autoantibodies compared with the general population.

    Finally, we did not include genetic factors in the analyses. Although results vary, genetic variants are suggested to explain some difference between patient outcomes41 ,42 and reported to associate with autoantibody positivity.43 ,44 To what extent genetic factors could have a modifying effect on the association between autoantibodies and the risk of mortality is not known.

    A strength of our study is that we included three large independent observational studies with data collected during the same period. All cohorts were initiated in the 1990s. Treatment was then less aggressive than today, this may have resulted in a higher mortality rate compared with patients treated with nowadays regimens.45 Furthermore, all cohorts had a long-term follow-up. As a result of the high number of patient-years of observation we were therefore able to differentiate the effects of RF, ACPA and anti-CarP individually.

    In conclusion, the present study showed that RF, independently of presence of other autoantibodies, is associated with increased all-cause mortality and with neoplasm and respiratory disease attributed death, and that ACPA is associated with CV death in RA.

    References

    1. ↵
      1. Bukhari M,
      2. Lunt M,
      3. Harrison BJ, et al
      . Rheumatoid factor is the major predictor of increasing severity of radiographic erosions in rheumatoid arthritis: results from the Norfolk Arthritis Register Study, a large inception cohort. Arthritis Rheum 2002;46:906–12. doi:10.1002/art.10167
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Kroot EJ,
      2. de Jong BA,
      3. van Leeuwen MA, et al
      . The prognostic value of anti-cyclic citrullinated peptide antibody in patients with recent-onset rheumatoid arthritis. Arthritis Rheum 2000;43:1831–5. doi:10.1002/1529-0131(200008)43:8<1831::AID-ANR19>3.0.CO;2-6
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Mikuls TR,
      2. Saag KG,
      3. Criswell LA, et al
      . Mortality risk associated with rheumatoid arthritis in a prospective cohort of older women: results from the Iowa Women's Health Study. Ann Rheum Dis 2002;61:994–9. doi:10.1136/ard.61.11.994
      OpenUrlAbstract/FREE Full Text
    4. ↵
      1. Gonzalez A,
      2. Icen M,
      3. Kremers HM, et al
      . Mortality trends in rheumatoid arthritis: the role of rheumatoid factor. J Rheumatol 2008;35:1009–14.
      OpenUrlAbstract/FREE Full Text
    5. ↵
      1. Wallberg-Jonsson S,
      2. Ohman ML,
      3. Dahlqvist SR
      . Cardiovascular morbidity and mortality in patients with seropositive rheumatoid arthritis in Northern Sweden. J Rheumatol 1997;24:445–51.
      OpenUrlPubMedWeb of Science
    6. ↵
      1. Young A,
      2. Koduri G,
      3. Batley M, et al
      . Mortality in rheumatoid arthritis. Increased in the early course of disease, in ischaemic heart disease and in pulmonary fibrosis. Rheumatology (Oxford) 2007;46:350–7. doi:10.1093/rheumatology/kel253
      OpenUrlAbstract/FREE Full Text
    7. ↵
      1. Farragher TM,
      2. Goodson NJ,
      3. Naseem H, et al
      . Association of the HLA-DRB1 gene with premature death, particularly from cardiovascular disease, in patients with rheumatoid arthritis and inflammatory polyarthritis. Arthritis Rheum 2008;58:359–69. doi:10.1002/art.23149
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Sihvonen S,
      2. Korpela M,
      3. Mustila A, et al
      . The predictive value of rheumatoid factor isotypes, anti-cyclic citrullinated peptide antibodies, and antineutrophil cytoplasmic antibodies for mortality in patients with rheumatoid arthritis. J Rheumatol 2005;32:2089–94.
      OpenUrlAbstract/FREE Full Text
    9. ↵
      1. Kuller LH,
      2. Mackey RH,
      3. Walitt BT, et al
      . Determinants of mortality among postmenopausal women in the women's health initiative who report rheumatoid arthritis. Arthritis Rheum 2014;66:497–507. doi:10.1002/art.38268
      OpenUrl
    10. ↵
      1. Humphreys JH,
      2. van Nies JA,
      3. Chipping J, et al
      . Rheumatoid factor and anti-citrullinated protein antibody positivity, but not level, are associated with increased mortality in patients with rheumatoid arthritis: results from two large independent cohorts. Arthritis Res Ther 2014;16:483. doi:10.1186/s13075-014-0483-3
      OpenUrlCrossRefPubMed
    11. ↵
      1. Ajeganova S,
      2. Andersson ML,
      3. Frostegård J, et al
      . Disease factors in early rheumatoid arthritis are associated with differential risks for cardiovascular events and mortality depending on age at onset: a 10-year observational cohort study. J Rheumatol 2013;40:1958–66. doi:10.3899/jrheum.130365
      OpenUrlAbstract/FREE Full Text
    12. ↵
      1. López-Longo FJ,
      2. Oliver-Miñarro D,
      3. de la Torre I, et al
      . Association between anti-cyclic citrullinated peptide antibodies and ischemic heart disease in patients with rheumatoid arthritis. Arthritis Rheum 2009;61:419–24. doi:10.1002/art.24390
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Thorsteinsson J,
      2. Bjornsson OJ,
      3. Kolbeinsson A, et al
      . A population study of rheumatoid factor in Iceland. A 5-year follow-up of 50 women with rheumatoid factor (RF). Ann Clin Res 1975;7:183–94.
      OpenUrlPubMedWeb of Science
    14. ↵
      1. Jónsson T,
      2. Thorsteinsson J,
      3. Valdimarsson H
      . Rheumatoid factor isotypes and cancer prognosis. Cancer 1992;69:2160–5. doi:10.1002/1097-0142(19920415)69:8<2160::AID-CNCR2820690824>3.0.CO;2-Z
      OpenUrlCrossRefPubMed
    15. ↵
      1. Shi J,
      2. Knevel R,
      3. Suwannalai P, et al
      . Autoantibodies recognizing carbamylated proteins are present in sera of patients with rheumatoid arthritis and predict joint damage. Proc Natl Acad Sci USA 2011;108:17372–7. doi:10.1073/pnas.1114465108
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Shi J,
      2. van de Stadt LA,
      3. Levarht EW, et al
      . Anti-carbamylated protein (anti-CarP) antibodies precede the onset of rheumatoid arthritis. Ann Rheum Dis 2014;73:780–3. doi:10.1136/annrheumdis-2013-204154
      OpenUrlAbstract/FREE Full Text
    17. ↵
      1. Gan RW,
      2. Trouw LA,
      3. Shi J, et al
      . Anti-carbamylated protein antibodies are present prior to rheumatoid arthritis and are associated with its future diagnosis. J Rheumatol 2015;42:572–9. doi:10.3899/jrheum.140767
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. Brink M,
      2. Verheul MK,
      3. Rönnelid J, et al
      . Anti-carbamylated protein antibodies in the pre-symptomatic phase of rheumatoid arthritis, their relationship with multiple anti-citrulline peptide antibodies and association with radiological damage. Arthritis Res Ther 2015;17:25. doi:10.1186/s13075-015-0536-2
      OpenUrlCrossRefPubMed
    19. ↵
      1. Shi J,
      2. van de Stadt LA,
      3. Levarht EW, et al
      . Anti-carbamylated protein antibodies are present in arthralgia patients and predict the development of rheumatoid arthritis. Arthritis Rheum 2013;65:911–15. doi:10.1002/art.37830
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Arnett FC,
      2. Edworthy SM,
      3. Bloch DA, et al
      . The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31:315–24. doi:10.1002/art.1780310302
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      1. Svensson B,
      2. Schaufelberger C,
      3. Teleman A, et al
      . Remission and response to early treatment of RA assessed by the Disease Activity Score. BARFOT study group. Better Anti-rheumatic Farmacotherapy. Rheumatology (Oxford) 2000;39:1031–6. doi:10.1093/rheumatology/39.9.1031
      OpenUrlAbstract/FREE Full Text
    22. ↵
      1. Symmons DP,
      2. Barrett EM,
      3. Bankhead CR, et al
      . The incidence of rheumatoid arthritis in the United Kingdom: results from the Norfolk Arthritis Register. Br J Rheumatol 1994;33:735–9. doi:10.1093/rheumatology/33.8.735
      OpenUrlAbstract/FREE Full Text
    23. ↵
      1. de Rooy DP,
      2. van der Linden MP,
      3. Knevel R, et al
      . Predicting arthritis outcomes–what can be learned from the Leiden Early Arthritis Clinic? Rheumatology (Oxford) 2011;50:93–100. doi:10.1093/rheumatology/keq230
      OpenUrlAbstract/FREE Full Text
    24. ↵
      1. van der Helm-van Mil AH,
      2. Knevel R,
      3. van der Heijde D, et al
      . How to avoid phenotypic misclassification in using joint destruction as an outcome measure for rheumatoid arthritis? Rheumatology (Oxford) 2010;49:1429–35. doi:10.1093/rheumatology/keq013
      OpenUrlAbstract/FREE Full Text
    25. ↵
      1. Barbarroja N,
      2. Pérez-Sanchez C,
      3. Ruiz-Limon P, et al
      . Anticyclic citrullinated protein antibodies are implicated in the development of cardiovascular disease in rheumatoid arthritis. Arterioscler Thromb Vasc Biol 2014;34:2706–16. doi:10.1161/ATVBAHA.114.304475
      OpenUrlAbstract/FREE Full Text
    26. ↵
      1. Hjeltnes G,
      2. Hollan I,
      3. Førre Ø
      , et al. Anti-CCP and RF IgM: predictors of impaired endothelial function in rheumatoid arthritis patients. Scand J Rheumatol 2011;40:422–7. doi:10.3109/03009742.2011.585350
      OpenUrlCrossRefPubMed
    27. ↵
      1. Gerli R,
      2. Bartoloni Bocci E,
      3. Sherer Y, et al
      . Association of anti-cyclic citrullinated peptide antibodies with subclinical atherosclerosis in patients with rheumatoid arthritis. Ann Rheum Dis 2008;67:724–5. doi:10.1136/ard.2007.073718
      OpenUrlFREE Full Text
    28. ↵
      1. Marasovic-Krstulovic D,
      2. Martinovic-Kaliterna D,
      3. Fabijanic D, et al
      . Are the anti-cyclic citrullinated peptide antibodies independent predictors of myocardial involvement in patients with active rheumatoid arthritis? Rheumatology (Oxford) 2011;50:1505–12. doi:10.1093/rheumatology/ker121
      OpenUrlAbstract/FREE Full Text
    29. ↵
      1. Arnab B,
      2. Biswadip G,
      3. Arindam P, et al
      . Anti-CCP antibody in patients with established rheumatoid arthritis: Does it predict adverse cardiovascular profile? J Cardiovasc Dis Res 2013;4:102–6. doi:10.1016/j.jcdr.2012.09.003
      OpenUrlCrossRefPubMed
    30. ↵
      1. Sokolove J,
      2. Brennan MJ,
      3. Sharpe O, et al
      . Brief report: citrullination within the atherosclerotic plaque: a potential target for the anti-citrullinated protein antibody response in rheumatoid arthritis. Arthitis Rheum 2013;65:1719–24. doi:10.1002/art.37961
      OpenUrlCrossRefPubMedWeb of Science
    31. ↵
      1. Cambridge G,
      2. Acharya J,
      3. Cooper JA, et al
      . Antibodies to citrullinated peptides and risk of coronary heart disease. Atherosclerosis 2013;228:243–6. doi:10.1016/j.atherosclerosis.2013.02.009
      OpenUrlCrossRefPubMed
    32. ↵
      1. Saksela E,
      2. Pyrhonen S,
      3. Timonen T, et al
      . Blocking effect of rheumatoid factor on the in vitro cytotoxicity of lymphoid cells from carcinoma patients. Scand J Immunol 1976;5:1075–80. doi:10.1111/j.1365-3083.1976.tb03059.x
      OpenUrlCrossRefPubMed
    33. ↵
      1. Pyrhönen S,
      2. Timonen T,
      3. Heikkinen A, et al
      . Rheumatoid factor as an indicator of serum blocking activity and tumour recurrences in bladder tumours. Eur J Cancer 1976;12:87–94. doi:10.1016/0014-2964(76)90210-3
      OpenUrlCrossRefPubMed
    34. ↵
      1. Giuliano AE,
      2. Irie R,
      3. Morton DL
      . Rheumatoid factor in melanoma patients: alterations of humoral tumor immunity in vitro. Cancer 1979;43:1624–9. doi:10.1002/1097-0142(197905)43:5<1624::AID-CNCR2820430510>3.0.CO;2-R
      OpenUrlCrossRefPubMed
    35. ↵
      1. Shmerling RH,
      2. Delbanco TL
      . The rheumatoid factor: an analysis of clinical utility. Am J Med 1991;91:528–34. doi:10.1016/0002-9343(91)90190-9
      OpenUrlCrossRefPubMedWeb of Science
    36. ↵
      1. Shi J,
      2. van Veelen PA,
      3. Mahler M, et al
      . Carbamylation and antibodies against carbamylated proteins in autoimmunity and other pathologies. Autoimmun Rev 2014;13:225–30. doi:10.1016/j.autrev.2013.10.008
      OpenUrlCrossRefPubMed
    37. ↵
      1. Wang Z,
      2. Nicholls SJ,
      3. Rodriguez ER, et al
      . Protein carbamylation links inflammation, smoking, uremia and atherogenesis. Nat Med 2007;13:1176–84. doi:10.1038/nm1637
      OpenUrlCrossRefPubMedWeb of Science
    38. ↵
      1. Apostolov EO,
      2. Basnakian AG,
      3. Ok E, et al
      . Carbamylated low-density lipoprotein: nontraditional risk factor for cardiovascular events in patients with chronic kidney disease. J Ren Nutr 2012;22:134–8. doi:10.1053/j.jrn.2011.10.023
      OpenUrlCrossRefPubMed
    39. ↵
      1. Berg AH,
      2. Drechsler C,
      3. Wenger J, et al
      . Carbamylation of serum albumin as a risk factor for mortality in patients with kidney failure. Sci Transl Med 2013;5:175ra29. doi:10.1126/scitranslmed.3005218
      OpenUrlAbstract/FREE Full Text
    40. ↵
      1. Speer T,
      2. Owala FO,
      3. Holy EW, et al
      . Carbamylated low-density lipoprotein induces endothelial dysfunction. Eur Heart J 2014;35:3021–32. doi:10.1093/eurheartj/ehu111
      OpenUrlAbstract/FREE Full Text
    41. ↵
      1. Viatte S,
      2. Plant D,
      3. Han B, et al
      . Association of HLA-DRB1 haplotypes with rheumatoid arthritis severity, mortality, and treatment response. JAMA 2015;313:1645–56. doi:10.1001/jama.2015.3435
      OpenUrlCrossRefPubMed
    42. ↵
      1. van Steenbergen HW,
      2. van Nies JA,
      3. Ruyssen-Witrand A, et al
      . IL2RA is associated with persistence of rheumatoid arthritis. Arthritis Res Ther 2015;17:244. doi:10.1186/s13075-015-0739-6
      OpenUrlCrossRefPubMed
    43. ↵
      1. Raychaudhuri S,
      2. Sandor C,
      3. Stahl EA, et al
      . Five amino acids in three HLA proteins explain most of the association between MHC and seropositive rheumatoid arthritis. Nat Genet 2012;44:291–6. doi:10.1038/ng.1076
      OpenUrlCrossRefPubMed
    44. ↵
      1. Farragher TM,
      2. Plant D,
      3. Flynn E, et al
      . Association of a rheumatoid arthritis susceptibility variant at the CCL21 locus with premature mortality in inflammatory polyarthritis patients. Arthritis Care Res 2010;62:676–82. doi:10.1002/acr.20208
      OpenUrlCrossRef
    45. ↵
      1. van Nies JA,
      2. de Jong Z,
      3. van der Helm-van Mil AH, et al
      . Improved treatment strategies reduce the increased mortality risk in early RA patients. Rheumatology (Oxford) 2010;49:2210–16. doi:10.1093/rheumatology/keq250
      OpenUrlAbstract/FREE Full Text
    View Abstract

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Handling editor Tore K Kvien

    • Contributors All authors were involved in drafting the manuscript or revising it critically for important intellectual content, and all authors approved the final version to be published. SA, JHH, TWJH, LAT, SMMV and AHMvH-vM contributed to concept of the study and data interpretation. SA, JHH and HWvS performed data analyses. SA, JHH, MKV, JABvN, IH, BS, LAT and SMMV contributed to data acquisition and AHMvH-vM supervised the study.

    • Funding We acknowledge the financial support from the Dutch Arthritis Foundation, The Netherlands Organization for Scientific Research, the IMI JU funded project BeTheCure, contract no 115142-2, the Swedish Rheumatism Association and King Gustav V 80 year's Foundation. Both AHMvH-vM and LAT are each supported by a ZON-MW Vidi grant. NOAR is supported by Arthritis Research UK, grant reference numbers 20380 and 20385. JHH is funded by Arthritis Research UK grant reference number 19743.

    • Competing interests TWJH and LAT are on a patent application for the use of anticarbamylated protein antibodies in diagnostics.

    • Patient consent Obtained.

    • Ethics approval The study was conducted with approval of the local institutional ethic committees at the Karolinska University Hospital, Lund University Hospital, Sahlgrenska University Hospital and Linköping University Hospital, Sweden (BARFOT); Norfolk and Norwich University Hospital ethics committee, UK (NOAR); and the Leiden University Medical Center, the Netherlands (EAC).

    • Provenance and peer review Not commissioned; externally peer reviewed.